Low back pain patients (n=142) on sick leave for at least 8 weeks were given a multi-modal cognitive behavioural treatment program (MMCBT) that lasted for 4 weeks. Before treatment, all patients were tested with a comprehensive test battery. The outcome at 12-month follow-up was predictable from the pretest, but only when combining medical and psychological data. Patients who returned to work (Returners, 50%) in the MMCBT group were characterised by less pain, more psychological strength, were evaluated by the physiotherapist as having a good prognosis for return to work, and were less educated. Patients who did not return to work (Non-returners) in the MMCBT group felt tense and unfit, felt hopelessness concerning the future, were less physically active, thought their complaints would worsen if they continued working, and reported fewer difficulties driving a car. Returners to work (58%) in the randomised control group (n=81), who received ordinary physical therapy, were characterised by high levels of energy, less subjective health complaints, less exhaustion for a condition test, and did not work in positions giving a constant load on the back. There was no significant differences between number of patients who had returned to work in the MMCBT and the control group. Non-returners in the control group lacked energy, trained less regularly, worked in occupations that gave an almost constant load on the back, and did not expect to be back to work in the course of a couple of weeks. It seems to be important to develop further diagnostic tools to identify those who might benefit from extensive or specific treatments. Copyright 1998 European Federation of Chapters of the International Association for the Study of Pain.